July 2015 doc.: IEEE 802.15-15-0500-00-007a Project: IEEE ... · doc.: IEEE 802.15-15-0500-00-007a...

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015

Roberts [Intel]Slide 1

Project: IEEE P802.15 Working Group for Wireless Personal Area Networks (WPANs)

Submission Title: LOS Link Budget

Date Submitted: July 2015

Source: Rick Roberts [Intel]

Address

Voice: 503-712-5012, E-Mail: [email protected]

Re:

Abstract:

Purpose:

Notice: This document has been prepared to assist the IEEE P802.15. It is offered as a basis for

discussion and is not binding on the contributing individual(s) or organization(s). The material in this

document is subject to change in form and content after further study. The contributor(s) reserve(s) the right

to add, amend or withdraw material contained herein.

Release: The contributor acknowledges and accepts that this contribution becomes the property of IEEE

and may be made publicly available by P802.15.

mailto:[email protected]

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


This document is a revision of work originally presented to IEEE802.15.7 in September 2009.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


This contribution is in response to the call for channel models and address

the line-of-sight OCC channel.

Intel believes that for many OCC usages, a channel model is not needed

because there is no implied guaranteed quality of service since the light

source is a signal of opportunity. If performance is not adequate then the

user needs to move closer to the source to improve the signal-to-noise

ratio. However, the automotive use case is an exception (has a QoS

requirement) and will be given additional emphasis in the presentation.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


ToC

• Radiometric (Physical) vs. Photometric (Visual)

• Path loss due to line-of-sight (LOS) light propagation

• Beam Divergence

• Atmospheric Attenuation Due to Fog

• Propagation Path Loss

• 850 nm NIR specific analysis

• Appendix A: Ascertaining the LED parameters of

interest

• Appendix B: Calculating integrated spectral flex

density

• Appendix C: Receiver noise density calculations

• Appendix D: Solid angle path loss model

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Radiometric (Physical) vs. Photometric (Visual)

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


LED Detector

Diode

Eyeball

A white LED spews

optical power across a

range of wavelengths

(mW/Hz)

The human eye and the detector diode have different frequency responses and hence

perceive the same LED source differently.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


For data link budgets we want to use Radiometric units

For illumination applications we want to use Photometric units (which include the

frequency response of the human eye)

Most VIS LED vendors generally only provide Photometric data since illumination

is the market today and the use of VIS LEDs for data is an obscure usage.

Radiometric

(Physical)

Photometric

(Visual)

Total Flux Watts (W) lumens (lm)

Flux Density W/cm2 lm/cm2

Source Intensity W/sr candela = lm/sr

Illuminance Lux (lx) = lm/m2

Irradiance W/m2

Units

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Best way to determine optical power and spectrum

… measure it using a known aperture sensor!

Often IR LED vendors will provide radiometric data for LEDs intended for

communications. Appendix A discusses a method to convert photometric units

to radiometric units, but it is not recommended by the author. It is felt that the

better method is to obtain radiometric data by measurement.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Path loss due to line-of-sight (LOS)

light propagation

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


An image sensor is an array of photodiodes behind a lens.

A single photodiode behind a lens can be considered an

image sensor with just one pixel. Thus, the analysis of this

degenerate case, to a first degree approximation, serves to

analyze each photodiode in an image sensor.

Spherical Light Source

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


In either case the total received flux (watts) of the

light source with respect to the tangential normal

within the field of view is to be determined.

Planer Light Source

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


When determining the tangential normal transmitted spectral flux density (watts per unit

area per wavelength W/m2nm), there are obviously two cases to consider: 1) area of the

source exceeds the FOV; 2) FOV exceeds the area of the source.

Case 1: Use the average flex density within the FOV.

Case 2: Use the total available flex since the total source area is within the FOV.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


The area of the aperture opening determines the amount of light entering the camera.

The total flex entering the camera is the product of the aperture area and the

irradiance flex density.

http://electronics.howstuffworks.com/cameras-photography/tips/aperture1.htm

http://electronics.howstuffworks.com/cameras-photography/tips/aperture1.htm

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


-+

+Vshutter/aperture

PD

The detector diode

current is integrated to

provide the energy per

bit in Joules.

A=C/sC=J/V

J

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


300 n

m 800 nmTotal

Current

Light

Source

Spectrum

Photodiode

Responsivity

The total current is given as the integral

over wavelength of the source spectrum

weighted by the photodiode wavelength

dependent responsivity. Appendix B

provides the mathematical details.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Beam Divergence

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Beam Divergence

Spherical Source Planar Source

D1 D2

Surface power density is inversely

proportional to the surface area

increase: see slide 20 for details.

Preferably, the light intensity

divergence should be supplied by the

vendor. The alternative is to

measure the divergence and then use

the estimates presented on slide 22.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Camera aperture on diverged surface power density

Spherical Source Planar Source

D1 D2

The camera aperture is projected on to the increased surface

area, of which the surface area has reduced power density.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Spherical Bulb Beam Divergence

D1 D2

𝐷𝑖𝑠𝑝𝑒𝑟𝑠𝑖𝑜𝑛 𝐿𝑜𝑠𝑠 = 10 ∙ 𝑙𝑜𝑔10𝐴2𝐴1

= 20 ∙ 𝑙𝑜𝑔10𝐷2𝐷1

For a spherical bulb, the dispersion loss is proportional to the increase

in distance squared (similar to RF with an isotropic radiator).

For a spherical source, all illuminated pixels are orthonormal since

they fall within the field of view.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Light Panel Beam Divergence

A1A2

Squaring both sides

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Observations on panel light dispersion:

1. dispersion is proportional to dispersion angle

2. dispersion increases as distance squared

3. dispersion is inversely proportional to the size of the panel

4. the area ratio 1.0

𝐷𝑖𝑠𝑝𝑒𝑟𝑠𝑖𝑜𝑛 𝐿𝑜𝑠𝑠 = 10 ∙ 𝑙𝑜𝑔10𝐴2𝐴1

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


The light panel source case can also experience

additional loss due to camera viewing angle.

A precise analysis would require vendor data on the light panel angular radiation.

Lacking such data, an approximation can be made as

Angular Loss ≈ 10 ∙ 𝑙𝑜𝑔10 𝑐𝑜𝑠 ∅ .

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Atmospheric Attenuation Due to Fog

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Comparison of laser beam propagation at 785 nm and 1550 nm in fog and haze for optical wireless communications;

Isaac I. Kim, Bruce McArthur, and Eric Korevaar; www.ece.mcmaster.ca/~hranilovic/woc/resources/local/spie2000b.pdf

From the paper by Kim, et. al. …

Λ 𝑑𝐵𝑘𝑚 = 10 ∙ 𝑙𝑜𝑔10 𝑒

3.91𝑉

𝜆550 𝑛𝑚

−𝑞

where V = visibility in km

= wavelength in nm

q = the size distribution of the scattering particles

= 1.6 for high visibility (V > 50 km)

= 1.3 for average visibility (6 km < V < 50 km)

= 0.16 V + 0.34 for haze visibility (1 km < V < 6 km)

= V – 0.5 for mist visibility (0.5 km < V < 1 km)

= 0 for fog visibility (V < 0.5 km)

For the most part, the distances of interest are much less than a kilometer so we can

express this on a per meter basis as

Λ 𝑑𝐵𝑚 = 0.01 ∙ 𝑙𝑜𝑔10 𝑒

3.91𝑉

𝜆550 𝑛𝑚

−𝑞

http://www.ece.mcmaster.ca/~hranilovic/woc/resources/local/spie2000b.pdf

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


clearhazefog

Fog and Haze Attenuation by Wavelength

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


heavy fog light fog

Fog and Haze Attenuation by Wavelength

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Propagation Path Loss

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Propagation Loss (dB) = Dispersion Loss (dB) + Angular Loss (dB) + (dB/m) distance (m)

Ingested Flex (W) = Source Flex Density Propagation Loss (ratio) Aperture Size

The signal-to-noise ratio (SNR) is defined as the ratio of the ingested flex to the

receiver noise. The receiver noise can be either calculated or measured. Given

real hardware, it is probably easier to measure the noise than calculate it since

such calculations would require extensive knowledge of the receiver structure.

Nevertheless, appendix C outlines a method of doing the calculations.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


850 nm NIR specific analysis

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


IEC60825 NIR Safety Limits

850 nm

9.64 W⋅cm-2⋅sr-1

Example of Class 1 laser device – e.g. laser pointer Note: 9.64 W⋅cm-2⋅sr-1 is bright … but SUN radiance is 2.4 kW⋅cm-2⋅sr-1

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


ATX = 1 cm2

Adet

Interpretation of IEC60825 NIR Safety Limits

=1 det2

detdet, 1 AD

D

ATX

D

To a first order approximation …

… how much power is ingested by the detector?

WD

AAsr

srcm

WcmAP TXTXTX

2

detdet,2

2

det 6.96.9

Interpretation for a ATX = 1 cm2

2

det

D

ALchannel WAP TXTX 6.9

See appendix D for more discussion

on source intensity using solid angles.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Receiver noise floor is a mixture of thermal noise and shot noise.

)()()( fSfSfS thermalshottotal

The biggest contributor to shot noise is the ambient solar spectral irradiance on a bright sunny day.

Receiver Noise

Sun spectral irradiance at 850 nm (sea level): 90 uW/cm2/nm

Short noise is caused by the random arrival time of photons from a light source.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Impact of Fog Attenuation

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Dense Fog Thick Fog Moderate Fog

DD

ALLL mdBfogdBchanneldBtotal

2

detlog10

0.01

0.1

1

10

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Att

en

ua

tio

n d

B/m

Visibility (V) km

850 nm Attenuation Per Meter in Fog

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Reaction Time and Stopping Distance ImplicationsAssumptions:

- road coefficient of friction: 0.8

- reaction time: 2.5 secs (typical)

- road grade: level

0

50

100

150

200

250

0 20 40 60 80 100 120 140 160

Vis

ibili

ty D

ista

nce

me

ters

km/hr

Visibility Distance vs. Speed

active cruise control range*

*http://www.bmw.com/com/en/insights/technology/technology_guide/articles/active_cruise_control_stop_go.html

http://easycalculation.com/engineering/civil/vehicle-stopping-distance.php

Automotive Industry

Dilemma: what to do

when the sensor

technology out performs

the human eye? Do we

encourage blind driving?

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Example Link Analysis ATX = cm2Adet = cm2

D

A detailed 850 nm link budget will be presented

as part of the Intel proposal to IEEE802.15.7r1.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Appendices

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Appendix A

Ascertaining the LED parameters of interest

On the following pages, equation (2.2.1) and Figure 2.1.1 are from the book

Introduction to Solid-State Lighting by A. Zukauskas, et.al. The equation

relates the power spectral distribution S() (W/nm) to luminous flux v (lm).

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


V() is the relative luminous efficiency function defined by CIE and given in the table

(from internet) and curve (from the book Fig 2.1.1)

780

380

683 ( ) ( )

nm

t t

nm

F S V d The LED total luminous flux Ft (lumens) is given as (2.2.1)

Sometimes it is convenient to use a Gaussian curve fitting for V() (from internet)

2285.4( 0.559)( ) 1.019 : inV e m ,

Find transmitted power and spectral density

Fig 2.1.1

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


'( ) ( )t t tS c S

Typically we only know a normalized spectral curve St’() instead of St() in (2.2.1). Denote

their relation as St()=ct St’() with an unknown scaling factor ct that can be found from

780

'

380

683 ( ) ( )

tt nm

t

nm

Fc

S V d

Warm White

Neutral White

Cool White

780

'

380

683 ( ) ( )

nm

t t t

nm

F c S V d

Remark:

The above step to find St() can be

skipped if St() can be either

measured using a spectrometer or

supplied by the LED vendor.

780

380

( )

nm

t t

nm

P S d

H

L

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


A normalized spatial luminous intensity distribution gt() is provided by a vendor.

We need to find the axial intensity I0 that is defined as the luminous intensity

(candelas) on the axis of the source (zero solid angle). Since the luminous flux Ft

is also a spatial integral of spatial luminous intensity in addition to spectral integral

we used before, we have the following relationmax max

0 0

0 0

* ( ) 2 ( )sint t tF I g d I g d

max0

0

2 ( )sin

t

t

FI

g d

where max and max are the source beam solid

angle and maximum half angle respectively and

max=2(1-cosmax).

normalized spatial luminous intensity distribution

Find transmitter luminous spatial intensity distribution I0 gt()

Note: if the axial intensity is provided by

the vendor then one only need convert

the intensity from candelas to watts/sr.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Appendix B

Calculating Integrated Spectral Flex Density

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Flu

x /

ba

nd

wid

th (

W/n

m)

LED Spectrum

(white LED)

2

21150

250

)()()( RdLRSP

nm

nm

RX

Where T() is the transmitter power spectral density (W/nm)

R() is the detector responsivity (A/W at )

L() is the propagation loss (loss at )

The detector diode vendors are

giving us the info we needThis is the information we need

from the LED vendor

For best performance we want the detector spectral responsivity to be “matched” to

the LED spectral density. In general this is hard to due, especially for white LEDs.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Appendix C

Receiver Noise Density Calculations

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Determining the noise density No

What are the sources that contribute to the noise density?

Photodetector Noise

Transimpedance Amplifier Noise

Ambient “in-band” noise

Interference from other VLC sources

Others?

Ndiode

Ntransamp

Nambient

NVLC

Modulation Domain Spectrum

Modulation

Domain

Spectrum

Analyzer

http://upload.wikimedia.org/wikipedia/commons/5/5b/Leuchtstofflampen-chtaube050409.jpg

http://upload.wikimedia.org/wikipedia/commons/5/5b/Leuchtstofflampen-chtaube050409.jpg

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Ambient “In-Band” Noise Floor

This probably has to be empirically

measured for many different environments

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Interference from other VLC sources

For OCC this source of noise is unlikely because of

the angle-of-arrival mapping of the lens because the

interferer would have to be on the same angular

vector as the desired noise source. In general, an

interfering source will form an image elsewhere on

the image sensor and can be spatially filtered out.

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


CL

R2

2 2 2 ( ) 4shotnpd thermal D S B SHi i i q I I I kT R

q is the electron charge (1.6e-19 coulombs)

ID is the dark current

IS is the signal current

IB is the background light induced current

B is the bandwidth (B=1 Hz for N0)

k is Boltzmann’s constant (1.38e-23 J/K)

T is the Kelvin temperature (~290 K)

RSH is the shunt resistance

The detector itself contributes a noise density Ndiode (W/ Hz)

(A/ Hz)

R2C2

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


TI OPA111

R2=10e6 ohms

Transimpedance

Amplifier Noise Analysis

Ref. TI/Burr-Brown Application Bulletin SBOA060

“Noise Anaysis of FET Transimpedance Amplifiers”

Assume R3=0

http://focus.ti.com/lit/an/sboa060/sboa060.pdf

http://focus.ti.com/lit/an/sboa060/sboa060.pdf

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


The resistors and capacitors form

critical corner frequencies as shown

below:

2

2 2

1

2f

R C

2121 ||||2

1

CCRRfa

TI OPA111

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Typically op-amps have three noise regions … the above noise regions are for the TI

OPA111 op-amp. It is anticipated most outdoor VLC implementations will be

bandpass systems operating in noise region 3.

Noise Regions for the

TI OPA228

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


The approximation output noise is given by

niRn NNNN 00

3

00

2

2

12

3

3

0 1

C

CKN n

20 4kTRN R

where

2

2

22

2

22

0 4)(2 RRkTIIIqiRiiN SHBSDnopnpdnop

in

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Then electrical SNR is

( ) ( ) ( )rH

rL

sig r f DI S R R d

The signal current is given as

2

2

2

2

2

2

12

3

2

2

04)(241

/)()()(

RRkTIIIqikTRC

CK

RatedLRSR

N

E

SHBSDnop

b

rH

rL

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


2

2

2

2

2

2

12

3

2780

380

2

04)(241

/)()()(

RRkTIIIqikTRC

CK

RatedLRSR

N

E

SHBSDnop

nm

nmb

Hz

V

Hz

V 22

Hz

VsV

sA

J

s

sV

A

J

A

VK

K

J 2

Hz

V

A

V

Hz

A 222

Hz

VsVV

sC

CV

A

C

A

VAC

2222

2

2

/

Hz

V

sW

AW

A

V 22

1

1

Eb/No Dimensional Analysis

AVs

JW

sA

J

A

WV

s

CA

sA

J

A

V2

sHz

1

KT

KJk

coulombsCq

/

)(

Hz

V

AV

AV

KK

J 22

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Appendix D

Solid angle path loss model

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


α

β

θmax

D

r

Transmitter

Receiver

LOS Link Model

The receiver distance to the source is D

The receiver aperture radius is r and receiver area is Ar

The angle between receiver normal and source-receiver line is

The angle between source beam axis and source-receiver line (viewing

angle) is

The solid angle of the receiver seen by the source is r

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


The luminous angular intensity of the source at the receiver direction is

I0gt(), and therefore the receiver ingested luminous flux Fr=I0gt()r.

The luminous path loss can be represented as

where r is the receiver solid angle which satisfies Arcos()D2r .

max max max

0

2

0

0 0 0

( ) ( ) ( ) cos

2 ( )sin 2 ( )sin 2 ( )sin

t r t r t rrL

t

t t t

I g g g AFL

FI g d g d D g d

Power path loss Lp can be proven equal to luminous path loss LL as follows:

Optical power can be written as

In LOS free space propagation, path loss is assumed independent of

wavelength. Power path loss can be represented as Lp=S2()/S1()=P2/P1.

Luminous flux is related to S() as , which is

linear with S()

Therefore, LL=F2/F1=Lp.

( )H

L

P S d

683 / ( ) ( )H

L

F lm W S V d

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Now that we have known the optical power loss due to LOS propagation,

we can obtain the received optical spectral density from transmitter optical

spectral density as

Suppose we use a photodiode detector to

receive the signal light. We can obtain the

electrical power of the signal as

( ) ( ) ( )r p t L tS L S L S

The detector diode vendors are

giving us the info we need

Received optical power , ( ) ( ) rH

rL

r o r fP S R d

Find received optical and electrical power

( )D

qR

hc

21150

,

250

( ) ( ) ( )

nm

r e r r D L

nm

P S R R d R

f

where Sr() is the received light power spectral density (W/nm)

Rr() is the receiver filter spectral response

RD() is the detector responsivity (A/W at )

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


ideal

typical

Exemplary Optical Filter Response

http://www.newport.com/images/webclickthru-EN/images/2226.gif

http://www.newport.com/images/webclickthru-EN/images/2226.gif

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Summary of key steps to obtain received optical

power and electrical power

• Calculate transmitter (source) optical power from given luminous

flux Ft (lumens) and normalized spectral curve St’()

• Find the transmitter axial intensity I0 from given luminous flux Ft

and luminous spatial intensity distribution gt()

• Find receiver ingested luminous flux Fr from receiver solid angle

and transmitter luminous spatial intensity distribution I0 gt()

• Find the luminous path loss LL from Ft and Fr

• Prove power path loss Lp is equal to luminous path loss LL from

which to find received optical power spectral curve Sr()

• Calculate received optical power and electrical power

doc.: IEEE 802.15-15-0500-00-007a

Submission

July 2015


Comment on RX aperture vs. magnification factor

It is the author’s opinion that the RX aperture determines the

“brightness” of an observed object and that the field of view

determines the magnification factor of an observed object. That

is, for a given aperture the magnification factor determines how

big an object appears but not how bright an object appears.

The observed brightness is solely a function of the aperture

size. It should be noted that the field of view, and hence the

magnification factor, is a function of the focal distance.

Same aperture lens

Long focal distance

Short focal distance

July 2015 doc.: IEEE 802.15-15-0500-00-007a Project: IEEE ... · doc.: IEEE 802.15-15-0500-00-007a...

Documents

Transcript of July 2015 doc.: IEEE 802.15-15-0500-00-007a Project: IEEE ... · doc.: IEEE 802.15-15-0500-00-007a...